Automated detection of human gait events from conventional videography

Abstract

Real-time detection of gait events play a vital role in movement dependent control applications such as rehabilitation for lower limb amputations. It also helps in determination of spatio-temporal and kinematic parameters. Gyroscopes, inertial sensors, magnetometers and foot sensors are popular in the detection of gait events. They need to be mounted carefully, or foot should be placed specifically on foot pressure during detection. This study presents a framework for automated detection of gait events from conventional videography using passive markers at Robotics And Machine Analytic Laboratory (RAMAN Lab). The proposed Passive maker based Gait event detection (PMGED) algorithm automatically detects heel strike (HS) and toe-off (TO); the timing of stance and swing phase; the number of the gait cycle. Ten healthy subjects are considered to evaluate the robustness and reliability of proposed algorithm. The method is comparable when evaluated against human expert detection.